Linux/Unix>> MNIST-Handwritten-Digits-Classication>> 返回
项目作者： Xenia101

项目描述 ：
A Simple Handwritten Digits Classication
高级语言： Python
项目地址: git://github.com/Xenia101/MNIST-Handwritten-Digits-Classication.git


A Simple MNIST Handwritten Digits Classication

Handwriting Identification Using MNIST Dataset

  1. train_pixels,train_list_values = mnist.train.next_batch(n_train_data) 
  2. test_pixels,test_list_of_values = mnist.test.next_batch(n_test_data) 
  4. train_pixel_tensor = tf.placeholder("float",[None,784])
  5. test_pixel_tensor = tf.placeholder("float",[784])
  7. distance = tf.reduce_sum(tf.abs(tf.add(train_pixel_tensor,tf.negative(test_pixel_tensor))),reduction_indices=1)

Example

  1. predicted number :  4 => answer :  4
  2. predicted number :  3 => answer :  8
  3. predicted number :  6 => answer :  6
  4. predicted number :  6 => answer :  3
  5. predicted number :  1 => answer :  1
  6. predicted number :  3 => answer :  5
  7. predicted number :  5 => answer :  5
  8. predicted number :  2 => answer :  2
  9. predicted number :  0 => answer :  0
  10. predicted number :  1 => answer :  4
  11. Accuracy =  0.6

Execution / Test Environment

  • Windows 10 or Ubuntu Linux
  • Python 3.6

Usage

  • Input : ./mnist_data

    python3 Digits Classication.py

  • Output : Accuracy